A MAGNETIC RESONANCE IMAGING INVESTIGATION INTO

THE FUNCTION OF THE DEEP CERVICAL FLEXORS DURING

THE PERFORMANCE OF CRANIOCERVICAL FLEXION

Barbara Cagnie, PT, PhD,a Roseline D'Hooge, PT,b Eric Achten, MD, PhD,c

Dirk Cambier, PT, PhD,d and Lieven Danneels, PT, PhDd

a Post-doctoraSciences and Phy

b Researcher,Physiotherapy, G

c Professor, DHospital, Ghent U

d Professor, Dsiotherapy, Ghen

Submit requeDepartment of ReUniversity, De Pi(e-mail: barbara.

Paper submitt2010; accepted F

0161-4754/$3Copyright © 2doi:10.1016/j.

286

ABSTRACT

Objective: Evidence suggests that the deep cervical flexors (DCFs) are important for the control of the cervical spine.The craniocervical flexion (CCF) test is a clinical test developed for patients with neck pain disorders based on theaction of the DCFs. Because these muscles are deeply situated, it is difficult to reach the DCFs with surfaceelectromyography. Magnetic resonance imaging (MRI) can be used to measure these muscles in cross section. Theobjective of this study was (1) to determine the reliability of MRI for measuring cross-sectional area (CSA) of thelongus colli (Lco) and longus capitis (Lca) and (2) to evaluate the changes in CSA during contraction.Methods: Thirty healthy subjects aged 29 ± 9.3 years were imaged using MRI. The CSA of the Lco and Lca wasevaluated at 4 different levels (C0-C1, C2-C3, C4-C5, and C6-C7) at rest and during CCF.Results: The intraclass correlation coefficients for the CSA of the Lco and Lca showed good to excellent reliability(0.73-0.92), except at the C4-C5 level. There was a significant increase in CSA of both Lco (F = 6.79, P = .015) andLca (F = 19.20, P ≤ .001) due to CCF, and this was at different levels. The highest increases in CSA occurred at theC0-C1 level for the Lca (11.1%) and at the C2-C3 level for the Lco (17.4%).Conclusions: This study demonstrated that the action of CCF resulted in a contraction of the Lco and Lca atdifferent levels. The results indicate that MRI is a promising technique to evaluate changes in CSA duringcontraction. (J Manipulative Physiol Ther 2010;33:286-291)

Key Indexing Terms: Neck Muscles; Magnetic Resonance Imaging; Muscle Contraction

In recent years, research has focused on identifying andquantifying deficits in the deep cervical flexor (DCF)muscles in patients with neck pain disorders.1-5

Evidence suggests that these muscles are important forthe control of the cervical lordosis and maintenance ofcervical spine postural form.6-10 Jull and coworkers11

developed the craniocervical flexion test (CCFT) inresponse to the clinical need for more directed and specific

l Researcher, Department of Rehabilitationsiotherapy, Ghent University, Ghent, Belgium.Department of Rehabilitation Sciences andhent University, Ghent, Belgium.epartment of Radiology, Ghent Universityniversity, Ghent, Belgium.

epartment of Rehabilitation Sciences and Phy-t University, Ghent, Belgium.sts for reprints to: Barbara Cagnie, PT, PhD,habilitation Sciences and Physiotherapy, Ghentntelaan 185, 3B3, 9000 Ghent, Belgiumcagnie@ugent.be).ed January 5, 2010; in revised form February 5,ebruary 5, 2010.6.00010 by National University of Health Sciences.jmpt.2010.03.010

evaluation and therapeutic exercise for patients with neckpain disorders based on the muscles' primary anatomicalaction. Craniocervical flexion is the primary action of thelongus capitis (Lca) muscle that attaches to the cranium andthe superior portion of the longus colli (Lco) that attaches tothe first cervical vertebrae.7,10,11 In contrast, superficialcervical flexor muscles such as the sternocleidomastoidmuscles are not prime movers of CCF and are structurallymore suited to assist in flexing the lower cervical spine onthe thorax.

Because these muscles are deeply situated, it is difficultto reach the DCF with surface electromyography (EMG).Nevertheless, Falla et al12 described a novel surface EMGtechnique for the detection of DCF muscle activity.However, this technique is quite invasive and is not ableto differentiate between the Lco and Lca.13,14 A recentstudy has indicated the need to differentiate between theLco and Lca because there is a clear difference in activationof both muscles.15

Quantitative measurements of paraspinal muscle actioncan be obtained with both real-time ultrasonography (US)and magnetic resonance imaging (MRI), and there isgrowing support for their use in investigations of patientswith spinal pain.16-19

287Cagnie et alJournal of Manipulative and Physiological TherapeuticsMRI Study of Deep Neck Flexors FunctionVolume 33, Number 4

The validity and reliability of US to evaluate the cross-sectional area (CSA) of the Lco have recently beeninvestigated in different studies.20-22 Evaluation of CSA hasbeen shown to have questionable reliability that may be due toboth anatomical characteristics and methodologicallimitations.20 The small size of the muscle and difficulties inoutlining the boundaries of the muscle may explain the largevariability.During contraction, a significant, although limited,increase in muscle CSA was found. In contrast, Javanshir etal21 found more reliable values; however, they also evaluatedthe anterior posterior and lateral dimensions and performedthe measurement at a lower level (C6). To get more insightinto the function of both Lco and Lca, it is recommended toevaluate changes in CSA at different cervical levels.

Magnetic resonance imaging has the potential to provide amore reliable and objective assessment of these muscles, bothat rest and during CCF. Hides et al19 recently investigated,through the use of MRI, the activation of the transversesabdominis during the drawing-in action, which is a clinicaltest for this muscle. To the best of our knowledge, no studieshave investigated the changes in CSA of the Lco and Lcaduring the performance of the CCFT using MRI.

Therefore, the purposes of this study were (1) todetermine the reliability of MRI for measuring CSA ofthe Lco and Lca separately at different cervical levels and(2) to evaluate the changes in CSA of the Lco and Lcaduring contraction.

METHODS

SubjectsThirty volunteers (21 women and 9 men) with a mean

age of 29 ± 9.3 years participated in the study. Exclusioncriteria were recent neck pain, back pain, or headache fromcervical origin (b3 months) and contraindications to MRI (acardiac pacemaker, claustrophobia, implanted metals,unremovable piercings, aneurysm clips, carotid arteryvascular clamp, neurostimulator, cochlear or ear implants,and [possible] pregnancy within the first 3 months). Afterreceiving verbal and written information, each subjectsigned a consent form containing information about thenature of the study. This study was approved by the localEthics Committee of the Ghent University Hospital.

ProtocolSubjects were instructed in the action of CCF before

undergoing MRI. They were asked to perform a gentlenodding action by first flexing the occiput and thencontinuing lower down the neck, with the sternocleido-mastoid muscle relaxed and without lifting the head offthe surface. Subjects were thoroughly trained to performthe CCF as accurately as possible because it was notpossible to provide visual feedback once the subject wasinside the scanner.

After this training period, subjects were placed in theMRI. They were asked to repeat the CCF task 4 times, asaxial images were separately obtained at 4 different cervicallevels (C0-C1 level, C2-C3 level, C4-C5 level, C6-C7level). The reason for choosing these levels was as follows:The C0-C1 and C6-C7 levels were chosen because at theselevels it was easy to separate, respectively, the Lca and Lco.In previous research with EMG, the C2-C3 level waschosen to evaluate the activity of the DCFs because the Lcohas the greatest CSA at this level.3,12 The C4-C5 level wasthe level right in between the C2-C3 and C6-C7 levels.

The scans were performed in a fixed order (from the C0-C1 level to the C6-C7 level); and for each level, ameasurement at rest was first taken, followed by a sequenceduring contraction. During the task, participants were askedto obtain full contraction and to accurately maintain thislevel of contraction for 25 seconds during which the imagewas acquired. There was 2 minutes of rest between eachrepetition of the task.

Magnetic Resonance ImagingMagnetic resonance imaging was performed on a 3-T

magnet (MAGNETOM Trio-Tim System; Siemens AG,Erlangen, Germany) in combination with imaging matrixcoils. A flexible neck coil, 20 × 50 cm, fixed over theanterior aspect of the participant's neck was combined withthe phased-array spine coil as a receiver coil combination.

The subjects were placed in a comfortable and relaxedsupine position, with their hips flexed to 45° and legssupported by foam wedges. The head was positioned in aneutral position, without rotation, lateral flexion, orexaggerated lordosis. A sagittal localizing sequence wasfirst performed to identify cervical disk space intervals.Axial images parallel to the consecutive intervertebral diskswith a slice thickness of 5.5 mm were obtained at the 4different cervical levels.

A 2-dimensional FLASH gradient echo was taken usingthe following imaging sequences: field of view read, 160mm; relaxation time, 60 milliseconds; echo time, 4.92milliseconds; voxel size, 0.5 × 0.5 × 5.5 mm. Totalacquisition time per scan was 23 seconds.

Data ManagementImages from MRI were transferred to an independent

workstation (Leonardo, Siemens) and postprocessed withstandard software (Syngo VB13, Siemens). Measurementsincluded CSA of the Lca (C0-C1, C2-C3, and C4-C5 level)and Lco (C2-C3, C4-C5 and C6-C7 level) at rest and oncontraction (left and right) (Figs 1 and 2).

Measurements conducted on MRI images were per-formed by 3 independent operators who were blinded to theother results to determine reliability.

Fig 1. Magnetic resonance image at the C0-C1 level showing theCSA of the Lca (A) at rest and (B) during contraction. (Colorversion of figure is available online.)

Fig 2. Magnetic resonance image at the C2-C3 level showing theCSA of the Lca and Lco (A) at rest and (B) during contraction.(Color version of figure is available online.)

288 Journal of Manipulative and Physiological TherapeuticsCagnie et alMay 2010MRI Study of Deep Neck Flexors Function

Statistical AnalysisAll statistical analyses were performed using SPSS 15.0

for Windows (SPSS, Chicago, IL). The normality ofvariables was evaluated by the Kolmogorov-Smirnov test,which demonstrated a normal distribution (P N .05).

For analysis of intertester reliability, intraclass correla-tion coefficients (ICC2,1), standard error of measurement(SEM), and smallest detectable difference (SDD) were

used. Defined with respect to a 95% level of confidence, theSDD is equal to 1.96 √2⁎SEM. As the reliability at the C4-C5 level for both the Lco and Lca was very low, no furtheranalyses at this level were made.

A general linear model was used to investigate within-group differences in CSA for both the Lca and Lco for thefactors level (C0-C1 and C2-C3 for Lca and C2-C3 and C6-C7 for Lco), condition (rest vs contraction), and side of the

Table 1. Intraclass correlation coefficients, SEM, and SDD for themeasurements of CSA of the Lco and Lca at rest and during CCF

ICC (95% CI) SEM SDD

Lca C0-C1 Rest 0.92 (0.86-0.95) 0.10 0.27CCF 0.90 (0.82-0.94) 0.14 0.38

C2-C3 Rest 0.73 (0.49-0.86) 0.06 0.16CCF 0.83 (0.68-0.91) 0.05 0.15

Lco C2-C3 Rest 0.91 (0.82-0.95) 0.05 0.14CCF 0.80 (0.66-0.87) 0.10 0.27

C6-C7 Rest 0.88 (0.77-0.94) 0.07 0.19CCF 0.91 (0.81-0.96) 0.08 0.22

CI, Confidence interval.

Table 2.Measurements of CSA (mean ± SD) of the Lco and Lca atrest and during contraction

Mean (±SD) P value

Lca C0-C1 Rest 1.35 (±0.34)CCF 1.50 (±0.43) b.001

C2-C3 Rest 0.54 (±0.11)CCF 0.59 (±0.13) .043

Lco C2-C3 Rest 0.69 (±0.17)CCF 0.81 (±0.22) .031

C6-C7 Rest 1.11 (±0.20)CCF 1.18 (±0.26) .036

289Cagnie et alJournal of Manipulative and Physiological TherapeuticsMRI Study of Deep Neck Flexors FunctionVolume 33, Number 4

body (left and right). Significant main or interaction effectswere further evaluated by examining mean values and 95%confidence intervals for each level in each condition.Statistical significance was accepted at the .05 α level.

RESULTS

ReliabilityThe ICCs for the interrater agreement for the CSA of the

Lco and Lca ranged from 0.73 to 0.92 depending on themuscle evaluated, indicating a good to excellent reliability(Table 1). Low values of the SEM were found for bothmuscles (6.3%-12.3%), providing evidence of high consis-tency for these variables.

Rest vs ContractionThe multivariate analysis of variance yielded no

significant interaction effect between the different factors.There were significant main effects for level (Lca: F =129.45, P ≤ .001; Lco: F = 61.21, P ≤ .001) and condition(Lca: F = 19.20, P ≤ .001; Lco: F = 6.79, P = .015).Because there were no significant main effects for side ofthe body, the values of the right and left sides were averagedfor further analyses. There was a significant increase inCSA of both Lco and Lca due to CCF, and this was atdifferent levels (Table 2). In terms of percentages, theincrease in CSA was, respectively, 11.1% and 9.3% for theLca at the C0-C1 and C2-C3 level and 17.4% and 6.3% forthe Lco at the C2-C3 and C6-C7 level.

DISCUSSION

This study on healthy, pain-free subjects demonstratedthat the action of CCF resulted in a contraction of the Lcoand Lca at different levels. The results indicate that MRI is apromising technique to evaluate changes in CSA duringcontraction. The reliability of the protocol was good toexcellent for both the Lco and Lca, except at the C4-C5level.23 Furthermore, good reliability was established asevidenced by low values of the within-subject SEM (range,

6.3%-12.3%), which allows to determine whether theobserved difference in scores on retesting were within themeasurement error. Compared with previous US studies inwhich the Lco was measured at both the C5-C6 level andthe C6 level, SEM values in the present study are quitelower, indicating higher consistency when using MRI.20,21

The poor reliability at the C4-C5 level for the Lco andLca may be due to the fact that it is difficult at this level todistinguish between both muscles because, at this level, theLco and Lca run parallel side by side.

This study confirms previous studies in which CCF hasbeen shown to elicit EMG activation of the deep cervicalmuscles.11-14 However, in these studies, Lco and Lca wereinvestigated as one entity and were captured at only onelevel. A recent study has indicated the need to differentiatebetween the Lco and Lca because there is a cleardifference in activation of both muscles.15 Based on theirattachments, both muscles may have a different anatomicalaction. The Lca attaches superiorly onto the cranium anddeeply to the front of the cervical spine with its mostinferior attachment to the C6 vertebra.7,24 The Lcoattaches inferiorly to the cranium and runs from theanterior tubercle of the atlas to the body of the T3 vertebra.It had multiple insertions on vertebrae between C1 and C6.Therefore, the primary anatomical action of Lca is flexionof the craniocervical junction, whereas the primary actionof Lco is flattening of the cervical lordosis. Although CSAchanged at different levels, the most important increases inCSA occurred at the C0-C1 level for the Lca and at theC2-C3 level for the Lco. The higher increases at the higherlevels may be related to the action of the CCF and theprimary anatomical function of both muscles. However, itmay not be excluded that, as the task was repeated in afixed order from the C0-C1 level to the C6-C7 level, thetask was executed less accurately when evaluating thelower levels because of residual fatigue.

The increase in CSA is comparable with the findings ofan US study, in which the CSA of the Lco was measuredat the C5-C6 level at rest and during contraction.20 In thisstudy, the CSA of the Lco during contraction wassignificantly higher compared with that at rest, with anenhancement of 12%. However, the amount of increase isnot comparable with other studies in which the lumbar and

Practical Applications

• In asymptomatic subjects, the action of CCFresulted in a contraction of the Lco and Lca atdifferent cervical levels.

• The use of MRI to evaluate the CSA of the Lcoand Lca separately, both at rest and duringcontraction, is reliable.

• The most important increases in CSA occurred atthe C0-C1 level for the Lca and at the C2-C3 levelfor the Lco.

290 Journal of Manipulative and Physiological TherapeuticsCagnie et alMay 2010MRI Study of Deep Neck Flexors Function

cervical multifidus muscles were investigated with USduring contraction.25-27 In these studies, an increase of upto 60% between rest and contraction was noted. Severaloptions may explain this difference. First, the size andanatomical structure of the Lco and Lca differ from thelumbar muscles. The small CSA of the DCF (about 1 cm2

compared with 2-7 cm2 for lumbar multifidus dependingon the level) may influence the results.17,25 Secondly, incontrast to our study, Lee et al measured contractionagainst loads, indicating that the tasks were probablyperformed at higher loads.25 More specifically, Lee et aldemonstrated that the thickness at 50% of maximumcontraction was significantly increased compared with thatat 25% of maximum contraction.25 It was not possible todetermine the exact intensity of the CCF task, but it isassumed that this is much lower than the intensities used inthe study of Lee et al Third, compared with other cervicalmuscle measurements, in previous studies, the thicknesswas measured instead of the CSA, which logically leads togreater percentages in increase. Lee et al25 demonstratedthat US had an acceptable validity for measuring thethickness of the cervical multifidus muscle but not formeasuring its CSA. Muscle thickness is defined as thelargest distance from the dorsal to the ventral boundary.We presume that thickness would have been a betterparameter to indicate contraction, as we clearly remarked achange in shape in both muscles, especially in the Lca.However, as these muscles are very small, it was notpossible to define the largest distance from the dorsal tothe ventral boundary.

LIMITATIONS

The present results must be viewed within the limitationsof the study. The task was performed in the scanner, sothere was no visual feedback if the patient performed thetask accurately. However, efforts were made to teach thesubjects the proper movement before undergoing MRI. Aspreviously mentioned, another potential limitation of thestudy is that the images were taken in sequential order (fromthe C0-C1 level to the C6-C7 level) with a time interval ofonly 2 minutes. It is possible that there was residual fatiguein the cervical flexors that may have influenced the results.It may be questioned whether the changes in CSA were dueto muscle activity or mechanical compression of the muscleresulting from movement of the cervical spine, as thecontraction was not performed in a static way. However, asprevious EMG studies have indicated that there is a strongrelationship between the amplitude of DCF muscleactivation and the different incremental stages of theCCFT, we may assume that the changes in CSA do reflectmuscle activity.12 Furthermore, it would have beeninteresting to look at differences between sex; but becauseof the relatively small number of subjects and the unequal

distribution of male and female (9:21), it was notappropriate to evaluate the effect of sex.

Future research is needed in individuals with a history ofneck pain and muscle impairment to establish the influenceof symptoms on muscle activation when performing CCF.

CONCLUSIONS

This study on normal, pain-free subjects demonstratedthat the action of CCF resulted in a contraction of theLco and Lca at different levels. The results indicate thatMRI is a promising technique to evaluate changes in CSAduring contraction.

ACKNOWLEDGMENT

The authors would like to thank Mrs Bieke Bruynooghefor her assistance with the collection of the data.

FUNDING SOURCES AND POTENTIAL CONFLICTS OF INTEREST

This study was supported by the Research Foundation-Flanders (FWO). No conflicts of interest were reported forthis study.

REFERENCES

1. Falla D, Jull G, Rainoldi A, Merletti R. Neck flexor musclefatigue is side specific in patients with unilateral neck pain.Eur J Pain 2004;8:71-7.

2. Falla D. Unravelling the complexity of muscle impairment inchronic neck pain. Man Ther 2004;9:125-33.

3. Falla DL, Jull GA, Hodges PW. Patients with neck paindemonstrate reduced electromyographic activity of the deepcervical flexor muscles during performance of the craniocer-vical flexion test. Spine 2004;29:2108-14.

4. Jull G, Kristjansson E, Dall'Alba P. Impairment in the cervicalflexors: a comparison of whiplash and insidious onset neckpain patients. Man Ther 2004;9:89-94.

5. O'leary S, Jull G, KimM, Vicenzino B. Cranio-cervical flexormuscle impairment at maximal, moderate, and low loads is afeature of neck pain. Man Ther 2007;12:34-9.

291Cagnie et alJournal of Manipulative and Physiological TherapeuticsMRI Study of Deep Neck Flexors FunctionVolume 33, Number 4

6. Conley MS, Meyer RA, Bloomberg JJ, Feeback DL, DudleyGA. Noninvasive analysis of human neck muscle function.Spine 1995;20:2505-12.

7. Vasavada AN, Li S, Delp SL. Influence of musclemorphometry and moment arms on the moment-generatingcapacity of human neck muscles. Spine 1998;23:412-22.

8. Boyd-Clark LC, Briggs CA, Galea MP. Muscle spindledistribution, morphology, and density in longus colli andmultifidus muscles of the cervical spine. Spine 2002;27:694-701.

9. Boyd-Clark LC, Briggs CA, Galea MP. Comparativehistochemical composition of muscle fibres in a pre-and apostvertebral muscle of the cervical spine. J Anat 2001;199:709-16.

10. Mayoux-Benhamou MA, Revel M, Vallqe C, Roudier R,Barbet JP, Bargy F. Longus colli has a postural function oncervical curvature. Surg Radiol Anat 1994;16:367-71.

11. Jull G, O'leary S, Falla D. Clinical assessment of the deepcervical flexor muscles: the craniocervical flexion test.J Manipulative Physiol Ther 2008;31:525-33.

12. Falla D, Jull G, Dall'Alba P, Rainoldi A, Merletti R. Anelectromyographic analysis of the deep cervical flexormuscles in performance of craniocervical flexion. Phys Ther2003;83:899-906.

13. O'leary S, Falla D, Jull G, Vicenzino B. Muscle specificity intests of cervical flexor muscle performance. J ElectromyogrKinesiol 2007;17:35-40.

14. Falla D, Jull G, O'leary S, Dall'Alba P. Further evaluation ofan EMG technique for assessment of the deep cervical flexormuscles. J Electromyogr Kinesiol 2006;16:621-8.

15. Cagnie B, Dickx N, Peeters I, et al. The use of functional MRIto evaluate cervical flexor activity during different cervicalflexion exercises. J Appl Physiol 2008;104:230-5.

16. Elliott J, Jull G, Noteboom JT, Galloway G. MRI study of thecross-sectional area for the cervical extensor musculature inpatients with persistent whiplash associated disorders (WAD).Man Ther 2008;13:258-65.

17. Stokes M, Hides J, Elliott J, Kiesel K, Hodges P.Rehabilitative ultrasound imaging of the posterior paraspinalmuscles. J Orthop Sports Phys Ther 2007;37:581-95.

18. Whittaker JL, Teyhen DS, Elliott JM, et al. Rehabilitativeultrasound imaging: understanding the technology and itsapplications. J Orthop Sports Phys Ther 2007;37:434-49.

19. Hides J, Wilson S, StantonW, et al. AnMRI investigation intothe function of the transversus abdominis muscle during“drawing-in” of the abdominal wall. Spine 2006;31:E175-8.

20. Cagnie B, Derese E, Vandamme L, Verstraete K, Cambier D,Danneels L. Validity and reliability of ultrasonography for thelongus colli in asymptomatic subjects. Man Ther 2009;14:421-6.

21. Javanshir K, Mohseni-Bandpei M, Rezasoltani A, Amiri M,Rahgozar M. Ultrasonography of longus colli muscle: Areliability study on healthy subjects and patients with chronicneck pain. J Bodywork Move Ther 2009, doi:10.1016/j.jbmt.2009.07.005.

22. Jesus FMR, Ferreira PH, Ferreira ML. Ultrasonographicmeasurement of neck muscle recruitment: a preliminaryinvestigation. J Manual Manip Ther 2008;16:89.

23. Shrout P. Measurement of reliability and agreement inpsychiatry. Statistical methods in medical research. StatMethods Med Res 1998;7:301-17.

24. Kamibayashi LK, Richmond FJ. Morphometry of human neckmuscles. Spine 1998;23:1314-23.

25. Lee JP, Tseng WY, Shau YW, Wang CL, Wang HK, WangSF. Measurement of segmental cervical multifidus contractionby ultrasonography in asymptomatic adults. Man Ther 2007;12:286-94.

26. Kiesel KB, Uhl TL, Underwood FB, Rodd DW, Nitz AJ.Measurement of lumbar multifidus muscle contraction withrehabilitative ultrasound imaging. Man Ther 2007;12:161-6.

27. McMeeken JM, Beith ID, Newham DJ, Milligan P, CritchleyDJ. The relationship between EMG and change in thickness oftransversus abdominis. Clin Biomech (Bristol, Avon) 2004;19:337-42.